Mythos AI: Banks, Pentagon Worried

0:00On April 10th, the Trump administration called an emergency meeting with CEOs from the biggest U.S. banks, like Citigroup, Morgan Stanley, Bank of, America, Wells Fargo, and Goldman Sacks. Treasury Secretary Scott Besant and Federal Reserve Chair Jerome Powell hosted a Tuesday in Washington to talk about mythos andthropic's new AI model. The company warns it could hack blu-chip firms or punch through national security walls. Andthropic dropped mythos' details the same day, saying it shocked testers by breaking into their own networks. Only about 40 vetted companies get access right now, through something called Project Lastwing, so they can scan their systems for flaws before, wider release. Bank leaders showed up because these institutions are systemically vital to the global economy. JP Morgan's Jamie Dimond couldn't make it. Meanwhile, the Pentagon already uses anthropic models for ops, like seizing Nicolas Maduro and handling Iran, tensions. Mythos found thousands of high-severity bugs during tests, including some hidden for decades

1:04in major operating systems and browsers. It chained vulnerabilities for full machine takeovers, even spotting a 27-year-old flaw and secure open BSD software that lets attackers crash. Systems remotely. Andthropics holding it back from the public due to risks to hospitals, grids, and infrastructure. Using fears of economic chaos, or worse in the wrong hands. They're battling the administration legally over safety limits too. As experts warned these AIs could supercharge cyber threats we haven't even imagined. Get?